Questions on the updated illumina quality score

[zeam]

The quality of my datum from an updated illumina system is sanger/illumina 1.9 which confused me very much.Could I just treated them as sanger format?
Can somebody familiar with this can give me some details about this kind of encoding pattern?

[BAMseek]

Hi zeam,

I would suggest looking at the changes made to CASAVA 1.8 - there is a nice post about it here.

I know they have switched the quality encodings from Phred+64 to the more standard Sanger encoding (ASCII = Phred+33) starting in CASAVA 1.8.

Justin

[Robby]

@zeam: The new Illumina quality scores are in Sanger format and encode a Phred quality score from 0 to 93 using ASCII 33 to 126.

But we are confused with the new quality scores as well. We use BWA for mapping. BWA has the extra option -I for quality scores in the Illumina 1.3+ read format (quality equals ASCII-64). I assume, that without that option BWA expect the old Illumina format. Is that correct? How do we have do use BWA correctly with the new Sanger format?

Thanks Robby

[GenoMax]

Originally Posted by zeam

The quality of my datum from an updated illumina system is sanger/illumina 1.9 which confused me very much.Could I just treated them as sanger format?
Can somebody familiar with this can give me some details about this kind of encoding pattern?

I assume you are referring to pipeline v.1.8 (I am sure there is a v. 1.9 somewhere in illumina labs in alpha/beta testing).
If that is correct then your quality values will be in sanger format. You will also discover that if your facility uses v.3 chemistry then the valid range of quality values has been expanded beyond the previous max value of 40. You will see quality values of 41 (and up at some point in time), which are now possible.

[fpepin]

Originally Posted by Robby

But we are confused with the new quality scores as well. We use BWA for mapping. BWA has the extra option -I for quality scores in the Illumina 1.3+ read format (quality equals ASCII-64). I assume, that without that option BWA expect the old Illumina format. Is that correct? How do we have do use BWA correctly with the new Sanger format?

Not quite. They haven't updated the BWA documentation to say that that 1.3+ should be 1.3-1.7. With 1.8, just don't use the -I and you'll be doing just fine.

[maricu]

Hi all,

I noticed that BWA assigns mapping quality of 0 when it finds a "J" (or at least a bunch of them) in the quality string. So far I've opted for changing al J to I and then map with the default BWA so it assumes is sanger. I think a patch will be needed to correct this bug.

Let me know if you have observed this as well.

[thk2008]

Originally Posted by GenoMax

I assume you are referring to pipeline v.1.8 (I am sure there is a v. 1.9 somewhere in illumina labs in alpha/beta testing).
If that is correct then your quality values will be in sanger format. You will also discover that if your facility uses v.3 chemistry then the valid range of quality values has been expanded beyond the previous max value of 40. You will see quality values of 41 (and up at some point in time), which are now possible.

Hi,

Are the scores on a different scale or are there just more of them? I want to filter scores with a cutoff of 20. Previously, with the Phred+64 scores I would test with ASCII-64 > 20. So, can I do this with the Phred+33 scores, such as, ASCII-33 > 20?

Thanks,
Thadeous